metadata

base_model: jbochi/madlad400-3b-mt
library_name: peft
license: apache-2.0
tags:
  - generated_from_trainer
model-index:
  - name: madlad400-finetuned-mbk-tpi
    results: []
language:
  - mbk
  - tpi
model_type: Translation
pipeline_tag: translation

madlad400-finetuned-mbk-tpi

This model is a fine-tuned version of jbochi/madlad400-3b-mt for translation from Malol to Tok Pisin.

Model details

Developed by: SIL Global
Finetuned from model: jbochi/madlad400-3b-mt
Model type: Translation
Source language: Malol (mbk)
Target language: Tok Pisin (tpi)
License: closed/private

Datasets

The model was trained on a parallel corpus of plain text files:

Malol:

Malol Scriptures
License: All rights reserved, Wycliffe Bible Translators. Used with permission.

Tok Pisin:

Tok Pisin back-translation
License: All rights reserved, Wycliffe Bible Translators. Used with permission.

Usage

You can use this model with the transformers library like this:

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("sil-ai/madlad400-finetuned-mbk-tpi")
model = AutoModelForSeq2SeqLM.from_pretrained("sil-ai/madlad400-finetuned-mbk-tpi")

inputs = tokenizer("Your input text here", return_tensors="pt")
outputs = model.generate(**inputs)
print(tokenizer.decode(outputs[0]))

madlad400-finetuned-mbk-tpi

This model is a fine-tuned version of jbochi/madlad400-3b-mt on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.1783
Chrf: 79.0009

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0005
train_batch_size: 4
eval_batch_size: 32
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Chrf
0.2957	7.7108	1600	0.2136	76.6433

Framework versions

PEFT 0.12.0
Transformers 4.44.2
Pytorch 2.4.1+cu124
Datasets 2.21.0
Tokenizers 0.19.1